Podcast and Content Discovery
Building discovery tools for podcasts and long-form content using search APIs.
The problem
Podcast and long-form audio content is notoriously hard to search within — most search engines index a show's title and description page but not the spoken content itself — so discovering an episode that actually covers a specific topic usually means relying on incomplete show notes rather than the real content.
How the workflow is built
Queries target show notes, episode titles, and transcript pages (where a podcast publishes one) since that's the text search engines can actually index, supplemented where possible by a dedicated podcast-transcript or audio-search service for shows that don't publish transcripts, since general web search alone under-covers the medium.
Example queries
"[topic]" podcast episode discussion"[guest name]" podcast interview appearance"[show name]" transcript "[topic]"
Pitfalls to watch for
- Show notes are frequently a poor proxy for what's actually discussed in an episode — a notes page can be thin, generic, or focused on sponsors rather than content, so search results built on notes alone will under-surface episodes that don't have detailed notes even if the content itself is highly relevant.
- Coverage is uneven across the medium: video podcasts on platforms with auto-generated captions are far more search-discoverable than audio-only shows without transcripts, creating a structural bias in what a search-driven discovery tool surfaces.
- Episode discovery has a real recency skew — search engines and the web generally favor newer content, so evergreen older episodes that are still highly relevant to a topic can be systematically underranked relative to more recent, less substantive coverage.